>The argument for (2) is that it leads to bad design. It may also be a
>little bit less efficient for future sophisticated browsers. But some
>people want this facility, so I guess we just have to accept this in
I do not think this sufficient justification. I want marked sections,
subdecl's, TIME, DATE, etc. elements and form types. I want to allow
native-language markup for experimental cases, but I have not seen
anyone say "some people want it, so we'll just have to accept it".
>3. Symbolic names for styles, but more powerful than CLASS and CSS1
> currently provide
>But for (3) there is a solution. In fact, in some of the earlier CSS1
>drafts this was already possible. However, the syntax for doing it was
>ambiguous, and some people objected to that. (The parser couldn't
>distinguish between a class name and an element name).
There are numerous ways of adressing elements in an SGML document, and
qualifying them based on the value of an attribute. Any one of these
could be used.